Evaluating Probabilistic Record Linkage for Reproductive Histories
Author Information
Author(s): Dorothea Nitsch, Susan Morton, Bianca L DeStavola, Heather Clark, David A Leon
Primary Institution: London School of Hygiene & Tropical Medicine
Hypothesis
How valid is probabilistic record linkage for identifying births in a historical cohort of women?
Conclusion
Probabilistic record linkage can reliably identify births in a population-based cohort with high specificity.
Supporting Evidence
- 68% of women responded to the questionnaire.
- 86% of respondents reported having had at least one birth.
- Linkage correctly identified 4930 births but missed 601 others.
- 134 births were incorrectly linked to the records.
Takeaway
This study looked at how well a method called probabilistic record linkage can match women to their birth records. It found that the method works pretty well.
Methodology
The study used probabilistic record linkage based on names, dates of birth, and postcodes to connect women from a cohort to their birth records.
Potential Biases
There may be bias due to non-response in the questionnaire, as only 68% of women returned it.
Limitations
The linkage missed some births, particularly those that occurred outside Scotland or before 1976 when the database coverage was incomplete.
Participant Demographics
The cohort consisted of women born in Aberdeen between 1950 and 1956, with a majority still residing in Scotland.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website